Overview

The AI Engineering with Claude Nanodegree program prepares developers to build and ship production-grade AI agent systems with Claude, the Claude Agent SDK, the Model Context Protocol (MCP), and Claude Code. Across four project-based courses, you'll go beyond demos to engineer the "harness" that runs an agent in production: controlling the agentic loop, governing MCP tools, evaluating systems so silent failures surface before customers do, and enforcing rules in code with bounded autonomy and human-in-the-loop guardrails. Portfolio projects are anchored in real industry scenarios from insurance, retail, financial services, and manufacturing, including an MCP-powered "Agentic Analyst" and an enterprise multi-agent code-review orchestrator.

Syllabus

Harness Engineering with Claude and Claude Code

Build the engineering skills to put Claude and Claude Code to work on real systems. You'll choose the right Claude model for a job by weighing intelligence, speed, and cost, then design agent architectures that perceive, reason, and act. Using the Claude Agent SDK, you'll build production agents driven by stop-reason loops and engineer context strategies that keep long conversations affordable. You'll configure Claude Code for a multi-surface monorepo team, author reusable Claude Skills, and stand up a multi-shift quality monitoring system with layered orchestration. A capstone ties it together: you run, verify, and defend the design of four working systems. Experience with Python, Typescript, and APIs is assumed.

MCP in Action

Connect AI applications to the tools, data, and systems they need using the Model Context Protocol (MCP). You'll start with the architecture: how hosts, clients, and servers talk to each other. From there you'll work through the core server features (tools, resources, and prompts) and client features like roots and sampling. You'll build your own MCP servers, integrate them with Claude Agent SDK agents, and add custom tools of your own. You'll also put governance in place with scoped configuration and an audited agent loop. By the end you'll build PriceScout, an agentic analyst that uses MCP tools to gather and reason over pricing data.

Agent Evaluation and Observability with Claude

Make AI agents you can trust by measuring what they do and watching how they behave in production. You'll enforce structured outputs with schemas so an agent's responses stay predictable and machine-readable. Then you'll build automated evaluation frameworks that score an agent's task completion, quality, and reliability. You'll apply these skills to real document-extraction systems: a validated, routed insurance policy pipeline and a resilient mortgage extraction system that holds up under messy input. You'll also fuse disagreeing evidence sources into a single supply-chain risk assessment. A final project brings evaluation and observability together.

Bounded Autonomy and Guardrails with Claude and Claude Code

Give AI agents room to act on their own while keeping them inside firm boundaries. You'll design multi-agent systems where an orchestrator coordinates specialized agents, then build a hub-and-spoke system with the Claude Python SDK. From there you'll add guardrails that hold: deterministic hooks that enforce agent compliance no matter what the model decides to do. You'll put it all to work building an enterprise code-review orchestrator that routes work across multiple agents and keeps each one accountable.